"the risk of upsetting tens to maybe hundreds of thousands of free users of its free distros is worth the possible benefit of making a considerably more robust OS which it can _sell_ to its big corporate clients?"
There are a lot of free distros out there. If their users didn't think Suse's were robust enough they'd jump ship. If they're the ones who make the decisions, or at least the recommendations, for the big corporate purchases and Suse alienates them they're not going to be on Suse's side when those purchasing decisions are made.
Adding what may well be perceived as a lot of weird, and therefore potentially flaky, stuff to something which is already perceived as good enough, is going to be a big point against buying it. And if the weird stuff really does prove to be flaky...
There's another point in the article which struck me, which I didn't comment on yesterday. If you're selling support and think that this improves robustness you're doing this to reduce your support costs. If, then, I'm a potential customer and look at what you're doing in that respect I'm likely to think if this is in the free edition and it's made it more robust* then why would I need to pay for the support?
Yet another alternative view is that if they system is designed to make rolling back upgrades easier if they go wrong, does it mean they're anticipating upgrades going wrong because they're planning to cut testing on them?
I suppose early encounters with Suse have left me feeling that it was a little outside the mainstream so I went Ubuntu > Debian > Devuan instead.
* This assumes I'm buying the "not robust enough to go without support" line.